Table 3.

Results of the 4-AFC human observer experiment with Monte Carlo mammograms of digital breast phantoms.

Reader #	1	2	3	4
${\mathrm{PC}}_{\mathrm{fulldose}}$ a	83.94 [79.52, 88.35]	83.94 [79.12, 88.35]	74.70 [69.08, 79.92]	75.90 [70.28, 81.12]
${\mathrm{PC}}_{\text{halfdose}}$	67.07 [61.45, 72.69]	67.07 [61.04, 72.69]	65.06 [59.04, 70.68]	65.06 [59.04, 70.69]
${\mathrm{PC}}_{\text{restored}}^{\mathrm{SAonly}}$	67.47 [61.85, 73.09]	65.46 [59.44, 71.08]	63.86 [57.83, 69.48]	66.27 [60.64, 71.89]
${\mathrm{PC}}_{\text{restored}}^{\mathrm{SA}+\mathrm{SP}}$	76.31 [71.08, 81.53]	72.29 [66.67, 77.91]	72.29 [66.67, 77.91]	71.89 [66.27, 77.11]
$\mathrm{\Delta }\mathrm{PC}$ (%)b	13.78	7.78	11.11	10.50
$p$-valuec	0.011	0.103	0.041	0.051

^aPC values and their 95% confidence intervals.

^bPC increase from ${\mathrm{PC}}_{\text{halfdose}}$ to ${\mathrm{PC}}_{\text{restored}}^{\mathrm{SA}+\mathrm{SP}}$.

^cProbability that there is no difference between ${\mathrm{PC}}_{\text{halfdose}}$ and ${\mathrm{PC}}_{\text{restored}}^{\mathrm{SA}+\mathrm{SP}}$.